Ai-Based Traffic Congestion Prediction for Smart Cities Using Artificial Neural Network

Abstract

As urbanization accelerates, efficient traffic management has become a critical challenge for smart cities. Traditional traffic prediction methods often struggle with the complexity and dynamic nature of city-wide congestion patterns. This study explores deep learning-based approaches for accurate and real-time traffic congestion forecasting. Using historical and real-time traffic data, we develop and evaluate neural network models (ANN) to capture spatiotemporal traffic dynamics. The proposed AI-driven framework integrates diverse urban data sources, such as road sensors, GPS trajectories, and weather conditions, to enhance predictive accuracy. Experimental results demonstrate that deep learning models outperform conventional statistical approaches in congestion prediction, offering valuable insights for traffic control, route optimization, and urban mobility planning. The findings highlight the potential of AI-powered traffic intelligence in developing smarter, more efficient, and sustainable urban transportation systems. When we evaluate using Kaggle datasets, we see that ANN model does better than other methods for precision, recall, F1-score as well as Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). Moreover the ANN model outperforms other methods in various time ranges. This comparison provides more evidence to support the effectiveness of this method for improving prediction accuracy in traffic congestion. It shows promise for a future where urban transport systems are smarter and more efficient